Philosophy of Mind Seminar (Friday - Week 2, MT25)

Abstract: As AI models become increasingly behaviourally sophisticated there is renewed interest in the extent to which we ought to give them moral consideration. If they could feel pain this would provide a strong basis for moral consideration, but many think that pain can only be felt by organisms (Sharkey 2025; Godfrey-Smith 2016) or that AI can only simulate or mimic cognition (Searle 1980, Schwitzgebel and Pober 2024). I will argue against these views. Considerations of evidential independence make it preferable to think that AI could meet a functional recasting of the criteria except for a putative biological criterion. I will argue that the scientific literature does not present sufficient empirical support for including a biological criterion and that the prominent philosophical arguments can be resisted. I will argue that the literature is too quick to rule out systems that simulate or mimic cognition. I will consider whether attributing pain to AI is morally reckless and conclude that these concerns are insufficient to overrule evidence of pain. On my view, we can adapt scientific tests for pain in animals to AI and in doing so we could find out whether they feel pain and deserve moral consideration.